You are on page 1of 10

ACI MATERIALS JOURNAL

TECHNICAL PAPER

MS No. M-2013-412.R2

Repeatability and Pervasiveness of Self-Healing in


Engineered Cementitious Composites
by Mustafa Sahmaran, Grkan Yildirim, Rezhin Noori, Erdogan Ozbay, and Mohamed Lachemi
This paper investigates the intrinsic self-healing ability of engineered cementitious composites (ECCs) coupled with multiple
microcrack formation under mechanical loading based on two
robustness criteria: repeatability and pervasiveness. To this end,
two different composites containing Class F fly ash and slag were
investigated. To generate microcracks, specimens were repeatedly preloaded up to 70% of their deformation capacities under
mechanical loading at the end of each specified cyclic wet/dry
conditioning period. Resonant frequency (RF) and rapid chloride
permeability tests (RCPT) were used to assess the extent of damage
and self-healing, and final results were supported by microscope
observations. RF measurements were recorded from two different
parts of each specimen (the top and middle portions) to monitor
whether self-healing takes place in certain regions or whether it
is pervasive over the entire specimen. Results of the experimental
study show that depending on the type of mineral admixture used
and the duration of initial curing before deterioration, ECC specimens can recover up to 85% of their initial RF measurements,
even after six repetitive preloading applications. The recovery
rates observed in the middle portion are similar to those in the top
portion for both ECC mixtures (to a slightly lesser extent), which
implies that self-healing is quite pervasive. Furthermore, after
repeated application of severe preloading, RCPT results for both
mixtures satisfy low or moderate chloride ion penetrability levels
in accordance with ASTM C1202. Due to the enhanced self-healing
capability of specimens, maximum crack width observed over the
specimen surfaces was restricted to 190 m (0.008 in.), even after
nine preloadings. These findings suggest that under certain conditions, the ECC materials produced in this study may significantly
enhance the functionality of structures by reducing the need for
repair and/or maintenance.
Keywords: engineered cementitious composite (ECC); repeatability; selfhealing; supplementary cementitious material (SCM).

INTRODUCTION
It is highly desirable that concrete material used in transport structures lasts for a long time without sacrificing its
mechanical and durability properties. However, due to a
number of deteriorating mechanisms, concrete unavoidably
cracks, which can lead to more severe degradation processes
and significant reductions in service life. Such damage
frequently requires repeated maintenance and/or repair of
degraded sections to compensate for lack of performance.
However, previous studies report that repairs are often
short-lived and that half of repairs applied in the field end
up failing.1 To make matters worse, continuous maintenance
and/or repair work is not always feasible, especially in the
case of large-scale concrete infrastructures, due to the significant time, labor, and capital required. In addition, owing to
ACI Materials Journal

structural restrictions and conditions, repairs can be difficult or


even impossible to execute in some cases. Under such circumstances, self-repair or self-healing of concrete material without
external human interference is an attractive alternative.
To date, many researchers have engaged in the study of
emerging techniques including the use of hollow fibers,2
chemical and bacterial encapsulation,3,4 expansive agents
and mineral admixtures,5,6 and intrinsic self-healing with
self-controlled tight crack width.7 Although all of the
approaches listed above hold promise, there are still a
number of concerns related to the robustness of the selfhealing mechanism.8 One such concern is the repeatability
of the self-healing phenomenon. Over the lifetime of an
infrastructure, damage is likely to occur more than once due
to multiple overloading and environmental conditions. Selfhealing capability should therefore be repeatable to satisfy
functionality, even in cases where crack opening occurs
from the same location as previous occurences. Another
concern related to self-healing robustness is the complexity
posed by different loading types on structures. Because it
is not always easy to predict cracking location, orientation,
and depth, healing that is pervasive rather than only present
in discrete parts of the structure is extremely appealing.
Although the aforementioned self-healing approaches have
limited effectiveness in fulfilling robustness criteria, the use
of intrinsic self-healing, coupled with the formation of many
closely spaced microcracks, seems to be the most promising
technique in situations where repeatability and pervasiveness of the self-healing mechanism are of concern.8 The
formation of many closely spaced microcracks is very difficult to achieve in conventional and fiber-reinforced concrete
(FRC) with the application of mechanical and environmental
loading. Therefore, the possibility of attaining consistent and
robust self-healing behavior with conventional concrete is
small. However, a newly developed material called engineered cementitious composites (ECC) has superior characteristics that outperform conventional concrete and FRC9,10
in terms of self-healing behavior.
ECC is part of a special category of high-performance
fiber-reinforced cementitious composites developed in the
last few decades. The optimization of ECC is based on the
application of micromechanics-based design theory to attain
ACI Materials Journal, V. 111, No. 1-6, January-December 2014.
MS No. M-2013-412.R2, doi: 10.14359/51687308, received May 19, 2014, and
reviewed under Institute publication policies. Copyright 2014, American Concrete
Institute. All rights reserved, including the making of copies unless permission is
obtained from the copyright proprietors. Pertinent discussion including authors
closure, if any, will be published ten months from this journals date if the discussion
is received within four months of the papers print publication.

Table 1ECC mixture proportions


Mixture
ID

Ingredients, kg/m3
PC

MA Water

F_ECC 566 680


S_ECC 593 712

PVA Sand HRWRA MA/PC W/CM

331

26

453

5.1

1.2

0.27

347

26

474

6.0

1.2

0.27

Note: 1 kg/m = 1.685 lb/yd .


3

Table 2Chemical and physical properties of


portland cement, fly ash, and slag
Chemical composition

PC

Fig. 1Typical flexural stress: midspan deflection curves of


ECC mixtures at 28 days.
high tensile ductility and tight microcrack formation under
direct tension at moderate fiber content (2% by volume, or
less).11 The extreme ductility of ECC is several hundred
times that of conventional concrete and FRC. Increased
ductility leads ECC materials to exhibit exceptional enhancement in toughness values, similar to that of metals.12 Along
with the improvement in toughness value, average crack
width remains at about 60 m (0.002 in.) regardless of ultimate tensile strain. The availability of information related
to this phenomenonm, especially in the case of multiple
overloading conditions, is lacking in the literature, as is the
answer to the question of whether self-healing occurs across
entire structures or is limited to certain areas. To account for
this knowledge gap, a detailed study was undertaken to gain
an understanding of the self-healing phenomenon observed
in ECC materials in terms of repeatability. To do so, 3- and
28-day ECC specimens incorporating different supplementary cementitious materials (SCMs) (Class-F fly ash and
slag) were produced. Then, on specified days, specimens
were subjected to repetitive preloading applications under
splitting tensile loading and subsequent environmental
conditioning to simulate self-healing behavior in cases where
crack formation takes place multiple times. Repeatability of
the self-healing mechanism was assessed by the use of rapid
chloride permeability (RCPT) and resonant frequency (RF)
tests. RF measurements were taken from different parts of
the specimens (top, bottom, and middle portions) to understand the extent of self-healing in different regions of specimens. Crack characteristics of repeatedly preloaded specimens were also evaluated based on the results obtained from
the surfaces of RF specimens.
RESEARCH SIGNIFICANCE
Although self-healing of cement-based materials is of
interest to many researchers, there is a lack of information related to this phenomenon, especially in the case of
multiple overloading conditions. The question of whether
self-healing takes place in the entire structure or is limited
to certain areas has also not been adequately answered in
previous studies. To account for this knowledge gap, an
experimental study was undertaken to examine the influence
of different SCMs and initial curing times on two robustness
criteria of self-healing: repeatability and pervasiveness.
2

CaO, %

61.43

3.48

35.09

SiO2, %

20.77

60.78

37.55

Al2O3, %

5.55

21.68

10.55

Fe2O3, %

3.35

5.48

0.28

MgO, %

2.49

1.71

7.92

SO3, %

2.49

0.34

2.95

K2O, %

0.77

1.95

1.07

Na2O, %

0.19

0.74

0.24

Loss on ignition

2.20

1.57

2.79

SiO2 + Al2O3 + Fe2O3, %

29.37

87.94

48.38

Physical properties
Specific gravity

3.06

2.10

2.79

Blaine fineness, m /kg

325

269

425

Note: 1 m /kg = 0.54 yd /lb.


2

EXPERIMENTAL PROGRAM
Materials and mixture proportions
Two different ECC mixtures incorporating Class F fly
ash (F_ECC) and ground-granulated blast-furnace slag
(S_ECC), which has the largest experimental dataset
published to date, were prepared. Figure 1 shows typical
stress-deflection graphs of both mixtures obtained under
four-point bending loading, which exhibit superior deflection-hardening behavior of the specimens at the end of 28
days. To be used for bending tests, prism specimens with
dimensions of 360 x 50 x 75 mm (14.2 x 1.9 x 2.9 in.) were
used and during the tests span length of flexural loading was
304 mm (12 in.) with a 101 mm (4 in.) center span length.
In both mixtures, the water-cementitious material ratio
(W/CM) and mineral admixture (fly ash [F] and slag [S])-portland cement ratio (MA/PC) were kept constant at 0.27 and 1.2,
respectively. The details of mixture proportions are shown
in Table 1. Both ECC mixtures contain CEM I-42.5 cement
(similar to ASTM Type I), fine silica sand with maximum
aggregate size of 400 m (0.016 in.), water, polyvinyl alcohol
fibers, and a polycarboxylic-ether type high-range waterreducing admixture (HRWRA) with a solid content of 40%.
Chemical composition and physical properties of cement, fly
ash, and slag are presented in Table2.
A series of cylindrical specimens with the dimensions of
150 x 300 mm (5.9 x 11.8 in.) was cast from a single batch
produced, using a 65 L capacity (0.085 yd3) gravity mixer
in resonant frequency (RF) tests. Specimens were demolded
after 24 hours and cured in plastic bags at 95 5% RH and
ACI Materials Journal

at 23 2C (73.4 3.6F) until the predetermined testing


ages. For rapid chloride permeability tests (RCPT), cylinder
specimens measuring 100 x 200 mm (3.9 x 7.9 in.) were
produced and kept in plastic bags at 95 5% RH and at 23
2C (73.4 3.6F). At the age of 28 days, cylindrical specimens with dimensions of 100 x 50 mm (3.9 x 2.0 in.) were
extracted with a diamond blade saw and used for RCPT tests.
Precracking and methods to evaluate repeatability
and pervasiveness of self-healing
Previous studies have shown that dynamic modulus
measurement based on ASTM C21513 and chloride ion penetrability measurement (RCPT) based on ASTM C120214
appear to be particularly promising and relatively simple
techniques for monitoring the extent and rate of autogenous healing.10,15 Although these test methods do not quantify damage, they provide useful ways of measuring the
extent and rate of self-healing without disturbing specimens in cases where healing was observed as a reduction
in material damage. For this reason, this study used these
test methods to observe self-healing rates under repetitive
preloading conditions. To monitor self-healing repeatability,
transverse RF measurements based on ASTM C21513 were
taken from 150 x 300 mm (5.9 x 11.8 in.) cylinder specimens. Before being exposed to cyclic preloading, four
3-day-old and four 28-day-old cylinders from both mixtures
were tested until failure under splitting tensile loading to
define ultimate deformation capacities of the ECC mixtures.
Ultimate splitting deformation capacities were in the range
of 2.0 to 2.6 mm (0.079 to 0.102 in.), in accordance with
mineral admixture type and initial curing period. The actual
tests were started after 3 and 28 days of initial curing, and
for each testing date, five cylindrical specimens were used
for RF measurements. Two were kept as virgin (untouched)
and the others were repeatedly preloaded up to 70% of
their original measured deformation capacities under splitting tensile loading to achieve various amounts of microcracking. Preloading was applied in each group of 10 cycles
up to 90 cycles for RF specimens. After each preloading, the
longitudinal sides of the specimens were sealed to restrict
water flow along lateral directions. Between each preloading
interval, specimens were subjected to wet-dry (W/D) cycles
consisting of submersion in water at 23 2C (73.4 3.6F)
for 24 hours and drying in a laboratory medium at 50 5%
RH and 23 2C (73.4 3.6F) for 24 hours. This type of
conditioning was selected with the intention of better simulating the actual field conditions since most of the structures in our daily lives are somehow subjected to actions of
wetting and drying. Because each W/D cycle represents a
period of 2 days, each 20-day interval corresponds with 10
complete W/D cycles. To determine whether self-healing is
restricted to certain regions of the specimens or is widely
dispersive over the entire area, RF measurements were taken
from four different points, two at the top and bottom and
two in the middle portions of the 150 x 300 mm (5.9 x 11.8
in.) cylinder specimens (Fig. 2). During the evaluation of
the RF test results, the average of the top and bottom point
measurements was accepted as the surface RF measurement,
and the average of the two middle point measurements as the
ACI Materials Journal

Fig. 2Schematic diagram of resonant frequency measurements at different depths. (Note: units in mm; 1 mm = 0.039 in.)
inside RF measurement. Additionally, changes in the widths
of cracks formed over the specimens were observed with the
help of microscope at the end of each 10 W/D cycles.
Repeatability of self-healing was also monitored through
RCPT tests. Ten 28-day-old cylindrical ECC specimens
with the dimensions of 100 x 50 mm (3.9 x 2.0 in.) were
tested from each mixture, based on ASTM C1202.14 Before
the application of preloading, the ultimate splitting deformation capacities of 100 x 50 mm (3.9 x 2.0 in.) cylinder
specimens after 28 days were found by taking the average
results of four specimens. The ultimate splitting deformation
capacities of F_ECC and S_ECC specimens were 1.7 mm
(0.067 in.) and 1.5 mm (0.059 in.), respectively. By taking
the final results into consideration, it was decided to keep
four of the specimens as virgin and to repeatedly preload the
remaining six specimens up to 70% of their maximum splitting deformation capacity under splitting tensile loading.
Preloaded specimens were then exposed to up to 100 W/D
cycles; RCPT measurements were recorded every 20 days.
RESULTS AND DISCUSSION
Resonant frequency (RF) test
Unhealed specimensResonant frequency (RF) test
results of ECC mixtures are presented in Fig. 3. In the figure,
measurements taken from the top and middle portions of
3- and 28-day virgin and preloaded ECC specimens are
illustrated. Results were achieved by using the initial RF
measurements and percentage variations observed beyond
3 and 28days until the end of 90 W/D cycles. When both
figures are carefully examined, it can clearly be seen that
there were changes in RF results depending on initial curing
period, SCM type, and number of W/D cycles. Original
RF values obtained from the top and middle portions of
the 3-day-old F_ECC specimens were 1870 and 1515 Hz,
respectively. In the case of 28-day-old F_ECC specimens,
the same values were 2050 and 2000 Hz. Original RF values
recorded from 3-day-old S_ECC specimens were 1925 and
1695 Hz for the measurements taken from top and middle
portion of the specimens, respectively. The same values
were 2070 and 2000 Hz in the 28-day-old S_ECC specimens. As seen from Fig. 3(a) and (e), an increasing trend
in RF results was monitored for 3-day-old virgin specimens
up to the end of 90 W/D cycles, irrespective of the point
where the RF reading was taken and the type of SCM used.
3

Fig. 3Percent variations in RF measurements due to repetitive preloading and subsequent conditioning.
The increase in RF results was similar for both the F_ECC
and S_ECC mixtures. For example, at the end of 90 W/D
cycles for 3-day-old virgin specimens, increments in RF
results reached up to 146% and 140% of the initial readings
for measurements taken from the top portions of F_ECC
and S_ECC specimens, respectively. The same modality
held true for readings taken from the middle portions, with
values reaching 129% and 121% for the same mixtures.
Although the final RF results of 3-day-old virgin ECC specimens incorporating fly ash and slag were similar, F_ECC
specimens showed slightly better performance than S_ECC
specimens in terms of changes in RF measurements. The
probable reason for this behavior in F_ECC specimens may
be due to pozzolanic reactions, which could result in greater
improvements in RF results compared with S_ECC specimens. This result implies that there is a higher probability
of F_ECC specimens consuming higher amounts of the
calcium hydroxide available in the hydrated matrix, especially at late ages, which could lead to greater improvements
in RF measurements compared to S_ECC specimens.
The difference in RF results was less pronounced for
virgin specimens initially cured for 28 days. As seen from
Fig. 3(b) and (f), after 90 W/D cycles, 28-day-old specimens
4

from both mixtures exceeded only 15% and 11% of the


initial RF results for readings taken from the top and middle
portions of specimens, respectively. It is also obvious from
the same figures that 20 W/D cycles were almost adequate for
complete stabilization of the results in the case of 28-day-old
ECC specimens. These specimens may have showed lower
enhancement in RF with increased initial curing time, due
to lower amounts of SCMs readily available to participate
in further hydration reactions and reductions in pore size,
densification of matrixes and consequently lower overall
transport of the moisture needed for further hydration.
Another point worth mentioning in the case of virgin specimens: regardless of initial curing time and type of SCM,
RF measurements taken from the top portion of specimens
were higher than those taken from middle portion at the end
of any number of W/D cycles. This behavior was observed
continually in nearly all of the specimens; the overall coefficient of variation was less than 7%, showing consistency
of results. The behavior is most probably attributable to the
changes in internal relative humidity of ECC specimens in
different regions. According to Jiang et al.,16 for cement
pastes with a W/CM higher than 0.40, moisture diffusion is
the only mechanism that leads to a reduction in internal relaACI Materials Journal

tive humidity; moisture diffusion and self-desiccation due to


chemical shrinkage of the paste have a simultaneous influence on the reduction of internal relative humidity in pastes
with a W/CM less than 0.40. Therefore, for ECC specimens
with a W/CM of 0.27, it can be stated that during the W/D
cycles, diffusion of moisture to the short distances (points
where the top and bottom RF measurements are taken) was
faster compared to longer distances (points where the middle
RF measurements are taken). This led to hydration kinetics
being faster and self-desiccation being counteracted more
easily at the top and bottom portion of specimens, which
resulted in higher RF measurements at the end of specified
testing cycles.
Effects of self-healingFigures 3(c), (d), (g), and (h) show
percentage changes in RF measurements taken from specimens repeatedly preloaded up to 70% of their maximum
deformation capacities under splitting tensile loading at the
end of each group of 10 W/D cycles until completion of
90 W/D cycles. Three specimens from each mixture were
initially cured for 3 and 28 days, and were then used to
investigate the likelihood of specimens exhibiting cracks in
varying numbers and widths at the end of each preload application. The averaged results are shown in these figures. As
the figures make it clear, the application of preloading and
exposure to different numbers of W/D cycles resulted in fluctuations with varying rates. After each preloading at the end
of each group of 10 W/D cycles, some reduction in RF results
was monitored. Results reflecting the self-healing effect
considerably changed depending on SCM type, number of
W/D cycles and initial curing time. As seen from Fig. 3(c),
(d), (g), and (h), the decrease in RF measurements with the
application of preloading showed significant recovery at the
end of different numbers of W/D cycles, although this trend
was not long-lasting. Careful analysis of RF measurements
from the top portion of 3-day-old ECC specimens shows
that reduced RF measurements with the applied preloadings
were not substantially recovered and recovery rates after
each period were similar beyond exposure to 60 W/D cycles.
The most probable reason for this equilibrium is the excessive widening of cracks upon preload repetitions and gradual
exhaustion of unhydrated cementitious materials at late ages
to account for crack closure through self-healing. However,
despite the stabilization of RF results beyond a certain time
period, after six preloading applications, average RF results
were only 85% and 74% of the initial RF measurement for
F_ECC and S_ECC mixtures, respectively. This shows that
the ECC materials produced in this study performed well
in terms of RF recovery under certain conditions, implying
the occurrence of self-healing, which could substantially
decrease repetitive repair and/or maintenance needs. As
stated above, the recovery of RF results just before stabilization was higher for F_ECC (85%) specimens than for
S_ECC (74%), and this tendency was observed to be true for
each RF measurement taken after each interval of 10 W/D
cycles. This result can be attributed to the changes in matrix
composition of ECC specimens. As previously mentioned,
the Class-F fly ash used in this study contains more SiO2,
which is important for pozzolanic activity, especially at later
ages, while slag contains higher amounts of CaO, which is
ACI Materials Journal

mostly responsible for self-cementing behavior. Therefore,


it is possible to conclude that S_ECC specimens performed
their cementing reactions earlier than F_ECC specimens,
and this led to faster consumption of unhydrated cementitious materials in S_ECC mixtures. F_ECC specimens
showed greater improvements in RF measurements after
each preloading and subsequent W/D cycles owing to the
abundance of unhydrated fly ash particles, even at later ages.
According to Sahmaran et al.,17 ECC specimens incorporating fly ash exhibited significantly tighter crack widths than
specimens with slag due to increased limitations in matrix
fracture toughness values. Thus, greater improvements in
RF results in F_ECC specimens could also be attributed to
the reduced crack widths occurring with the application of
cyclic preloading.
It is clear from Fig. 3(d) and (h) that the repeatability of
self-healing in terms of RF measurements decreased significantly in specimens initially cured for 28 days. As seen from
the figures, the recovery results of both ECC mixtures showed
stabilization after 30 W/D cycles, regardless of the point of
measurement. Average recoveries obtained from the top
portions were 80% and 71% of the initial RF measurements
for 28-day-old F_ECC and S_ECC specimens after three
preload repetitions and 30 W/D cycles, beyond which minimal
or no further RF recovery was monitored. The explanation for
this stabilization could be that due to increased initial curing
time, the matrix maturity of specimens enhanced significantly,
leaving fewer unhydrated cementitious materials available for
the further hydration reactions needed to prompt self-healing
behavior. Moreover, due to the continuity of hydration reactions, the matrix gets more compact at later ages, leading to an
increase in matrix-fiber frictional bond strength and fracture
toughness values. Increments in these parameters are detrimental to the attainment of sufficient multiple microcracking
and could trigger localized crack formations with wider
widths, which may negatively influence crack closure through
self-healing at later ages.17
As previously mentioned, two RF measurements were
taken from the surface and middle portions of the ECC specimens to better understand whether self-healing is limited
to certain regions of specimens with repeated preload applications and subsequent W/D conditioning, or whether it
is dispersive over the entire area. Generally speaking, RF
measurements from the middle portions of specimens were
lower compared to the recovery rates observed in the surface
portions (Fig. 3). This behavior was more explicit for 3-dayold F_ECC, and to a lesser extent, for 3-day-old S_ECC
mixtures. When the results from both surface and middle
portions of 3-day-old mixtures are compared, it is evident
that the number of recovered preload repetitions before the
stabilization of results did not change, although the extent of
recovery decreased in measurements taken from the middle
portions, regardless of mixture type. For example, surface
portion measurements taken after the application of six repetitive preloadings were 85% and 74% of recovery observed
in RF results for 3-day-old F_ECC and S_ECC mixtures,
respectively. The values were 66% and 67% for the same
mixtures in measurements taken from the middle portions.
The same modality was true for 28-day-old specimens, except
5

Table 3Rapid chloride permeability test results of ECC specimens


0 cycles
Mixture ID
Virgin
F_ECC
Preloaded
Virgin
S_ECC
Preloaded

Zero

AP

4154

4154

0 + 10 cycles

0 + 20 cycles

0 + 30 cycles

0 + 40 cycles

0 + 50 cycles

BP

AP

BP

AP

BP

AP

BP

AP

BP

AP

2673

1995

1995

1749

1749

1613

1613

1597

2673

(9.7)

(9.7)

(2.3)

(2.3)

(3.3)

(3.3)

(4.0)

(4.0)

(7.2)

(7.2)

(5.3)

4082

4925

3141

3611

2753

3279

2842

3440

3150

3626

3440

(4.6)

(6.7)

(5.8)

(7.5)

(8.1)

(7.0)

(1.7)

(3.2)

(2.7)

(9.6)

(3.2)

995

995

759

759

662

662

497

497

465

465

441

(5.5)

(5.5)

(4.9)

(4.9)

(0.53)

(0.53)

(2.0)

(2.0)

(7.3)

(7.3)

(5.0)

1064

1583

1160

1547

1302

1618

1409

1733

1639

1956

1885

(7.3)

(6.5)

(6.1)

(5.7)

(6.9)

(5.5)

(5.5)

(6.1)

(6.5)

(5.6)

(6.4)

*After preloading.

Before preloading.

Note: Numbers in parentheses are coefficient of variation (COV).

Fig. 4Percent variations in chloride ion permeability values of ECC mixtures due to repetitive preloading and subsequent conditioning.
for the fact that the number of recovered preload repetitions
decreased before the stabilization of RF results. In the case
of the 28-day-old specimens, after three repetitive preload
applications, RF results taken from the surface portions
started to stabilize at 80% and 71% levels for F_ECC and
S_ECC mixtures, respectively, while results from the middle
portions reached 71% and 67% for the same mixtures. It can
be deduced from these observations that near-surface cracks,
which are more easily exposed to water, can be healed up to
85% even after six bouts of severe deformation exposure due
to ease of moisture transport over shorter distances.
Rapid chloride permeability test (RCPT)
Unhealed specimensChloride ion permeability test
results of ECC specimens are tabulated in Table 3. All data
presented are averages of four specimens. The results are
expressed in terms of total electrical charge in Coulomb,
which reflects ECC materials ability to resist chloride ion
ingress. Because different SCMs show varying characteristics and low paste maturity at early ages, RCPTs were started
after 28 days of initial curing to account for higher-thannormal results.
According to the results obtained from virgin specimens
(Table 3), it is clear that undergoing 50 W/D cycles caused
marked reductions in RCPT results, regardless of SCM type.
Until the end of 50 W/D cycles, results showed a decreasing
trend between 4154 and 1597 and 995 and 441 Coulomb
in F_ECC and S_ECC mixtures, respectively. Another point
that is visible at first glance is that the drop in chloride ion
6

penetrability results of F_ECC specimens was significantly


higher than it was for S_ECC specimens. This behavior of
F_ECC specimens is more noticeable in Fig. 4, which shows
the percentage changes in chloride ion permeability results
with respect to the first measurement taken after 28 days of
initial curing versus number of W/D cycles. As seen from
the figure, the reduction in chloride ion charge/number of
cycle slope was found to be more explicit for F_ECC specimens than for S_ECC specimens, especially after the first
20 W/D cycles. For example, at the end of 20 W/D cycles,
average chloride ion penetrability results of F_ECC specimens dropped by 52%, while S_ECC specimens only
decreased by 33%. As the number of W/D cycles increased,
the chloride ion charge-number of cycle curve slope
smoothed out for both ECC mixtures due to gradual diminishment of unhydrated cementitious materials with time.
The probable reason for F_ECC virgin specimens exhibiting
notable improvement in chloride ion penetrability results
compared to S_ECC specimens could be due to the fact
that specimens incorporating slag have a greater quantity of
reacted particles in comparison to specimens incorporating
fly ash. This behavior of F_ECC specimens is also evident
in the fact that fly ash particles (especially Class-F fly ash)
remained untouched in the system up to 30 to 40% without
any chemical process,18 requiring longer periods to react
and produce hydration products such as calcium-silicatehydrate (C-S-H) gels through pozzolanic reactions. Although
the improvement in final values was more pronounced for
F_ECC specimens at the end of each W/D cycle, the chloACI Materials Journal

ride ion penetrability of S_ECC specimens was significantly


lower than in F_ECC specimens after any number of W/D
cycles. S_ECC specimens may have shown lower RCPT
results than F_ECC specimens due to finer particle size and
the higher cementing behavior of slag particles. Table 2
shows that slag (425 m2/kg [230 yd2/lb]) is significantly finer
than Class F fly ash (269 m2/kg [145 yd2/lb]). Therefore, it
can be deduced that the microstructure of S_ECC specimens
became more compact due to enhanced filler effect and the
higher cementing capability of finer slag particles, especially during the early stages of hydration. Because RCPT
is an electrochemical test method used for the evaluation of
chloride ion permeability, the electrical conductivity of the
pore solution should also be accounted for during the assessment of results. Electrical conductivity of the pore solution, which can be critical for chloride ion penetrability test
results, can be reduced significantly by lowering the concentration of alkali ions (Na+ and K+) and associated hydroxyl
ions (OH).19 Thus, the lower alkali content of slag particles may also be a reasonable explanation for the reduced
RCPT results of S_ECC specimens (see Table 2). It must be
emphasized that although the rate of improvement in RCPT
results was lower for virgin S_ECC mixtures, even without
the application of further W/D cycles, average chloride ion
permeability results of virgin S_ECC specimens were under
1000 Coulomb (995 Coulomb), regarded as the maximum
value by ASTM C120214 in terms of resistance to chloride
ion ingress. Virgin F_ECC specimens could never reach this
level of chloride transport properties.
Effects of self-healingChloride ion permeability test
results on repeatedly preloaded cylindrical (100 x 50 mm
[3.9 x 2.0 in.]) specimens are presented in Table 3. Data
provided in Table 3 was obtained by taking the average of
four specimens. With the application of repetitive preloading
and the cyclic W/D curing regime, results varied significantly. Figure 4 indicates that both ECC mixtures showed
self-healing, with decreasing chloride ion penetrability
results after each W/D cycle. The rate of self-healing for
S_ECC specimens was higher than for F_ECC specimens,
which was constant throughout the entire W/D cyclic exposure. However, the difference between the healing rates of
the ECC mixtures became less explicit due to a reduction
of unhydrated cementitious particles over time. Although it
is less prominent in the S_ECC specimens, the healing rate
for both mixtures decreased significantly beyond 30 W/D
cycles. This finding is in line with previously mentioned
conclusions about RF measurements taken from specimens
initially cured for 28 days.
Another striking point is that the application of five
repetitive preloading and 50 successive W/D cycles led
the average RCPT results of F_ECC specimens to decrease
from 4082 to 3440 Coulomb, and the results of S_ECC
specimens to increase from 1064 to 1885 Coulomb (Table
3). In other words, while the chloride ion penetrability test
results of S_ECC specimens increased up to 77% of the
initial reading after five preload applications and corresponding W/D cycles, F_ECC specimens decreased by
16% (Fig.4). Although comparable self-healing took place
in both mixtures, after each preloading and corresponding
ACI Materials Journal

cyclic W/D curing, the RCPT results of F_ECC specimens


showed a decreasing trend with respect to the first result
taken at the end of 28 days, while S_ECC specimens showed
the opposite. This behavior can be attributed to greater
fiber-matrix frictional bond strength and fracture toughness of S_ECC specimens compared to F_ECC specimens,
which may have significantly reduced the chance of multiple
crack occurrence and increased crack widths with the application of repetitive preloading. Moreover, S-ECC specimens
with an average 1064 Coulomb RCPT result after 28 days of
initial curing already showed very high resistance to chloride
ion penetration; beyond this time period, the possibility for
further improvements was significantly reduced. Therefore,
despite the self-healing performance of S_ECC specimens
observed after each number of W/D cycles, the formation of
cracks with wider widths (due to increased paste maturity)
and faster consumption of inner materials (due to enhanced
cementitious behavior) suppressed self-healing and caused
inadequate sealing of cracks, which led to a marked increase
in final RCPT results in S_ECC specimens (see Crack characteristics of preloaded specimens section). This finding is
a strong indication of the dependence of self-healing on tight
crack formation and availability of particles to be hydrated.
However, it is important to point out that under certain environmental conditions, even after five repetitive severe preload
applications, chloride ion penetrability results of both ECC
mixtures produced in this study (3440 and 1885 Coulombs for
F_ECC and S_ECC, respectively) satisfy the moderate or low
chloride ion penetrability levels prescribed by ASTM C1202.
Crack characteristics of preloaded specimens
The number of cracks and minimum, maximum, and
total crack opening measurements of 3- and 28-day-old
ECC specimens subjected to repetitive preloading and W/D
cycles are summarized in Table 4. Data presented in the table
represent average measurements recorded over two opposite
faces of three cylindrical specimens used for RF tests. As
seen from Table 4, despite significant self-healing of cracks
upon exposure to subsequent cyclic W/D conditioning, an
increase in the number of preload applications increased
both number of cracks and total crack widths to a certain
level. A higher number of cracks was observed in F_ECC
specimens than in S_ECC specimens. Although this characteristic of F_ECC specimens was observed to be true at the
end of each preload application and W/D cyclic exposure, it
was more explicit for specimens initially cured for 28 days.
The probable reasons that might lead F_ECC specimens to
show a higher number of cracks have already been explained
in previous sections and will not be further discussed here.
Figure 5 shows the rate of total crack closure observed
over entire crack surfaces in percentages, as well as the
number of cycles applied to 3- and 28-day-old specimens.
The values in Fig. 5 were calculated by the equation

PLTCW (t ) SH TCW (t +10)


PLTCW (t )

100

where PLTCW(t) and SHTCW(t+10) denote preloaded total crack


width at cycles t and self-healed total crack width at cycles
7

Table 4Crack characteristics of preloaded and self-healed ECC specimens


3-day-old
Preloaded
Mixture No. of
ID
cycles

F_ECC

S_ECC

Self-healed

Crack width, m

No. of
cracks

Min.

Max.

7 to 11

10

100

10

8 to 13

20

20

8 to 14

30
40
50

28-day-old
Preloaded

Crack width, m

Total

No. of
cracks

Min.

Max.

470

100

490

3 to 9

10

20

110

530

4 to 10

9 to 14

20

110

560

10 to 14

10

110

610

11 to 14

20

120

640

60

11 to 15

10

120

70

12 to 15

30

80

12 to 16

90

13 to 16

Self-healed

Crack width, m

Total

No. of
cracks

Min.

Max.

5 to 8

20

110

70

280

6 to 9

20

20

80

300

6 to 11

6 to 11

30

80

350

7 to 12

20

80

390

9 to 13

20

90

670

9 to 14

10

120

690

10 to 14

20

120

710

20

120

740

6 to 9

20

100

10

6 to 10

10

20

7 to 10

30

7 to 10

40

Crack width, m

Total

No. of
cracks

Min.

Max.

Total

520

110

550

3 to 6

20

80

390

10

120

580

4 to 7

20

90

450

7 to 11

20

120

630

5 to 9

30

90

520

8 to 11

20

130

710

7 to 10

10

110

600

475

9 to 12

20

130

750

7 to 11

20

110

680

100

550

9 to 12

30

130

790

8 to 11

20

120

720

20

110

590

10 to 12

20

130

810

9 to 12

30

120

760

11 to 14

20

100

660

10 to 13

20

140

830

10 to 12

10

120

790

11 to 15

10

100

690

11 to 13

30

140

860

10 to 12

20

130

810

560

4 to 7

20

150

540

100

530

3 to 6

10

60

280

5 to 8

30

150

570

3 to 5

20

110

310

20

110

550

4 to 8

20

60

310

6 to 8

30

150

630

4 to 6

20

120

370

20

110

580

5 to 8

10

70

370

6 to 9

40

150

740

5 to 7

10

120

570

8 to 12

20

120

680

6 to 9

20

90

460

7 to 9

30

160

780

6 to 8

30

130

680

50

9 to 12

30

120

720

7 to 10

30

110

580

8 to 9

20

160

810

7 to 9

20

130

720

60

10 to 12

30

140

880

8 to 11

10

110

650

8 to 9

30

180

820

7 to 9

30

150

770

70

11 to 13

30

150

920

9 to 12

20

110

820

8 to 9

40

180

890

8 to 9

20

170

800

80

11 to 13

20

150

1010

10 to 12

10

140

870

8 to 9

40

200

990

8 to 9

30

190

870

90

12 to 13

20

160

1070

10 to 12

10

150

980

8 to 9

40

200

1080

8 to 9

40

190

980

Note: 1 m = 0.00004 in.

Fig. 5Total crack closure rates of ECC specimens due to


self-healing.
t+10, respectively. When the crack closure rates of 3-day-old
ECC specimens exposed to cyclic preloading and subsequent
W/D conditioning are evaluated, it is evident that comparable
self-healing took place in both mixtures, although sealing
of cracks was slightly better in F_ECC specimens (Fig. 6).
For specimens initially cured for 28 days, however, the total
healing of cracks over the surfaces of S_ECC specimens
was more pronounced, especially at early ages. The extent
of self-healing monitored in the 28-day-old S_ECC specimens is rather surprising, since the crack widths were larger
8

than expected (Table 3), and significantly lower amounts


of unhydrated cementitious materials were expected to be
present in S_ECC matrixes, especially at later ages, due to
the enhanced cementing behavior of slag. The increased
self-healing capability of S_ECC specimens may be associated with the high pH value of the pore solution, which
may contribute to higher amounts of Ca2+ ions being leached
away from C-S-H gels and/or calcium hydroxide, leading to
enhanced CaCO3 growth and more evident final self-healing
behavior.7,20 Although the higher degree of self-healing was
observed in the 28-day-old S_ECC specimens, the final selfhealed condition of specimens was not as complete as that
of the F_ECC specimens; at the end of nine preloadings and
90 W/D cycles, total crack opening measured over specimen
surfaces was 810 m (0.032 in.) and 980 m (0.039 in.) for
F_ECC and S_ECC specimens, respectively. This suggests
that although self-healing kinetics are faster in bigger
cracks,21 the availability of required compounds and internal
water limit the extent, and small crack widths are crucial
for complete sealing. It is important to emphasize that under
certain environmental conditions, even after nine repetitive
preload applications, the maximum individual crack widths
observed in 28-day-old specimens were limited to 130 m
(0.006 in.) and 190 m (0.008 in.) for F_ECC and S_ECC,
respectively. These values are still below the minimum

ACI Materials Journal

Fig. 6Self-healing in the microcraks of 3-day-old ECC specimens.


threshold reported by different authors for the possibility of
complete sealing of a crack.22-24
Initial aging of specimens caused differences in repeatability of crack sealing performance. After nine repetitive
preloadings and cyclic W/D exposure, the cracks in 3-dayold specimens were healed from the same place four to five
times, while in the 28-day-old specimens, self-healing of the
same cracks was repeated two to three times, regardless of
SCM type. This situation was true for cracks with widths of
no more than 20 m (0.0008 in.) and 30 m (0.0012 in.) in F_
ECC and S_ECC specimens, respectively. Moreover, in cases
where crack widths were restricted to 20 to 30 m (0.0008
to 0.0012 in.), new cracks that formed next to healed cracks
were observed in most of the reloading situations. Although
these new cracks were close to former crack sites, due to the
availability of unhydrated cementitious materials in opposite
crack faces, they were able to heal rapidly. However, when
cracks widths were larger than approximately 100 m (0.004
in.), cracks followed the same paths that were healed initially.
Therefore, even under repetitive preloading, mechanical
recovery is possible, although further investigations on related
topics are needed for a more precise understanding.
CONCLUSIONS
This paper outlines a detailed study undertaken to gain
an understanding of the self-healing phenomenon observed
in ECC materials in terms of repeatability and pervasiness.
Based on the test results and analysis of self-healed ECC
specimens incorporating different SCMs (Class F fly ash and
slag), the following conclusions can be drawn:
Regardless of the point of measurement, for 3-day-old
ECC specimens, recovery of RF measurements after
the application of preloadings were not eye-catching
and were similar to each other after approximately six
preloadings and 60 W/D cycles were applied. In the
case of 28-day-old ECC specimens, however, stabilizaACI Materials Journal

tion in RF recovery results started after the application


of approximately three repetitive preload applications
and 30 W/D cycles. Therefore, depending on the type of
SCM used and initial curing time, the recovery rate in
RF measurements might reach up to 85% of the initial
measurement, even after six repetitive preloadings.
For measurements taken from the middle portions of
specimens, RF recovery rates were observed to be lower
compared to measurements taken from the surfaces of all
speciments, which can be attributed to the difficulty of
having moisture diffuse across longer distances. Although
RF recovery results dropped for the middle-portion
measurements, the number of recovered preload applications stayed the same as the top-portion measurements,
with slightly lower recovery rates. These findings show
that even under repetitive loading conditions, the selfhealing mechanism is widely dispersive over the entire
area of the specimens rather than being limited to specific
areas. However, the extent of self-healing is highly
dependent on easy exposure of cracks to water.
At the end of five repetitive preloading applications and
50 subsequent W/D cycles, RCPT results for F_ECC
specimens decreased up to 16%, while results for S_
ECC specimens increased up to 77% of the initial values,
despite comparable self-healing occurrence observed
in both mixtures. This behavior of S_ECC specimens
can be attributed to the formation of larger cracks and
faster consumption of unhydrated materials. While the
two mixtures behaved in different manners, final RCPT
results after five repetitive preloadings were 3440 and
1885 Coulombs for F_ECC and S_ECC mixtures,
respectively, which are defined as moderate and low
chloride ion penetrability levels in ASTM C1202.
Self-healing behavior considering the crack closure
performance was quite successful in the case of both
mixtures, although the extent was more pronounced for
9

specimens cured for limited periods (3 days). Enhanced


self-healing capability regarding with the closure of
cracks was also observable by checking maximum crack
widths over the specimens after self-healing so that even
after the application of severe preloading for nine times
followed by 90 W/D cycles, maximum crack widths
monitored over the specimens with no respect to the
age of specimens were 130 m (0.006 in.) and 190 m
(0.008in.) for F_ECC and S_ECC mixtures, respectively.
When ECC specimens were further aged, there were
reductions in self-healing rates in terms of RF and
RCPT measurements regardless of SCM type used in
the mixtures. This finding might look to be suggestive
in that self-healing is not effective when the specimens
were of higher maturity. However, it must be emphasized that likelihood for concrete material to be cracked
during early ages is higher due to sensitivity against
several shrinkage mechanisms, thermal movements and
so on. This implies that concrete is much more in need
of higher levels of self-healing to take place at the early
ages in comparison to later ages. Moreover, preloading
was repeated from the same point for nine repetitive
times in the present study, which is a rather rare situation to be observed in real-life structures. Overall, these
findings suggest that under certain environmental conditions, self-healing could be useful in real-life structures
no matter what the age is although more detailed studies
are needed on the related topic.
AUTHOR BIOS

Mustafa ahmaran is an Associate Professor in the Department of Civil


Engineering at Gazi University, Ankara, Turkey. His research interests
include advanced materials technology and composite materials development for sustainable infrastructures.
Grkan Yldrm is a PhD Student and Research Assistant in the Department of Civil Engineering at Gazi University. His research interests include
the development of composite materials for sustainable infrastructures.
Rezhin Noori is a Masters Student in the Department of Civil Engineering
at Gaziantep University, Gaziantep, Turkey. His research interests include
composite materials development for sustainable infrastructures.
Erdogan zbay is an Assistant Professor in the Civil Engineering Department at Mustafa Kemal University, Hatay, Turkey. His research interests
include durability of concrete, usage of waste materials in concrete, and
self-consolidating concrete.
ACI member Mohamed Lachemi is a Professor in the Department of Civil
Engineering at Ryerson University, Toronto, ON, Canada. His research
interests include the use of high-performance materials in the built infrastructure, including the development and use of self-consolidating concrete
in construction.

ACKNOWLEDGMENTS

The authors gratefully acknowledge the financial assistance of the Scientific and Technical Research Council (TUBITAK) of Turkey provided under
Project: MAG-112M876 and the Turkish Academy of Sciences Young
Scientist Award program.

REFERENCES

1. Mather, B., and Warner, J., Why Do Concrete Repairs Fail? interview held at University of Wisconsin, Department of Engineering Professional Development, Madison, WI, Nov. 5, 2004.
2. Mihashi, H.; Kaneko, Y.; Nishiwaki, T.; and Otsuka, K., Fundamental
Study on Development of Intelligent Concrete Characterized by Self-

10

Healing Capability for Strength, Japan Concrete Institute, V. 22, No.2,


2000, pp. 441-450.
3. Huang, H., and Ye, G., Application of Sodium Silicate Solution as
Self-Healing Agent in Cementitious Materials, International Conference
on Advances in Construction Materials through Science and Engineering,
2011, Hong Kong, pp. 530-536.
4. Jonkers, H. M., Bacteria-Based Self-Healing Concrete, HERON,
V.56, No. 1/2, 2011, pp. 1-12.
5. Kishi, T.; Ahn, T. H.; Hosoda, A.; and Takaoka, H., Self-Healing
Behavior by Cementitious Recrystallization of Cracked Concrete Incorporating Expansive Agent, First International Conference on Self-Healing
Materials, Apr. 18-20, 2007, Delft, the Netherlands.
6. Ahn, T. H., and Kishi, T., Crack Self-Healing Behavior of Cementitious
Composites Incorporating Various Mineral Admixtures, Advanced Concrete
Technology, V. 8, No. 2, 2010, pp. 171-186. doi: 10.3151/jact.8.171
7. ahmaran, M., and Li, V. C., Engineered Cementitious Composites: Can They Be Accepted as a Crack-Free Concrete? Transportation
Research Record, V. 2164, 2010, pp. 1-8. doi: 10.3141/2164-01
8. Li, V. C., and Herbert, E., Robust Self Healing Concrete for Sustainable Infrastructure, Advanced Concrete Technology, V. 10, No. 6, 2012,
pp. 207-218. doi: 10.3151/jact.10.207
9. Li, V. C., and Yang, E., Self Healing in Concrete Materials, Self Healing
Materials, Springer Netherlands, Dordrecht, the Netherlands, 2007, pp. 161-190.
10. Yang, Y.; Lepech, M. D.; Yang, E. H.; and Li, V. C., Autogenous
Healing of Engineered Cementitious Composites under Wet-Dry Cycles,
Cement and Concrete Research, V. 39, No. 5, 2009, pp. 382-390. doi:
10.1016/j.cemconres.2009.01.013
11. Li, V. C., Engineered Cementitious Composites Tailored Composites through Micromechanical Modeling, Fiber Reinforced Concrete:
Present and the Future, N. Banthia, A. Bentur, and A. Mufti, eds., Canadian
Society for Civil Engineering, Montreal, QC, Canada, 1997, pp. 64-97.
12. Maalej, M.; Hashida, T.; and Li, V. C., Effect of Fiber Volume Fraction on the Off-Crack Plane Fracture Energy in Strain-Hardening Engineered
Cementitious Composites, Journal of the American Ceramic Society, V.78,
No. 12, 1995, pp. 3369-3375. doi: 10.1111/j.1151-2916.1995.tb07979.x
13. ASTM C215-08, Standard Test Method for Fundamental Transverse, Longitudinal, and Torsional Frequencies of Concrete Specimens,
ASTM International, West Conshohocken, PA, 2008, 7 pp.
14. ASTM C1202-12, Standard Test Method for Electrical Indication
of Concretes Ability to Resist Chloride Ion Penetration, ASTM International, West Conshohocken, PA, 2012, 7 pp.
15. Jacobsen, S., and Sellevold, E. J., Self Healing of High Strength
Concrete after Deterioration by Freeze/Thaw, Cement and Concrete
Research, V. 26, No. 1, 1996, pp. 55-62. doi: 10.1016/0008-8846(95)00179-4
16. Jiang, Z.; Sun, Z.; and Wang, P., Internal Relative Humidity Distribution in High-Performance Cement Paste Due to Moisture Diffusion and
Self Desiccation, Cement and Concrete Research, V. 36, No. 2, 2006,
pp. 320-325. doi: 10.1016/j.cemconres.2005.07.006
17. Sahmaran, M.; Yucel, H. E.; Demirhan, S.; Arik, M. T.; and Li,
V.C., Combined Effect of Aggregate and Mineral Admixtures on Tensile
Ductility of Engineered Cementitious Composites, ACI Materials Journal,
V. 109, No. 6, Nov.-Dec. 2012, pp. 627-638.
18. Song, G., and van Zijl, G. P. A. G., Tailoring ECC for Commercial
Application, 6th International RILEM Symposium on Fiber Reinforced
Concrete, RILEM Publications SARL, Bagneux, France, 2004, pp. 1391-1400.
19. Shehata, M. H.; Thomas, M. D. A.; and Bleszynski, R. F., The Effect
of Fly Ash Composition on the Chemistry of Pore Solution in Hydrated
Cement Pastes, Cement and Concrete Research, V. 29, No. 12, 1999,
pp. 1915-1920. doi: 10.1016/S0008-8846(99)00190-8
20. Sahmaran, M.; Yildirim, G.; and Erdem, T. K., Self-Healing Capability of Cementitious Composites Incorporating Different Supplementary
Cementitious Materials, Cement and Concrete Composites, V. 35, No. 1,
2013, pp. 89-101. doi: 10.1016/j.cemconcomp.2012.08.013
21. Gagn, R., and Argouges, M., A Study of the Natural Self-Healing
of Mortars using Air-Flow Measurements, Materials and Structures, V. 45,
No. 11, 2012, pp. 1625-1638. doi: 10.1617/s11527-012-9861-y
22. Aldea, C.; Song, W.; Popovics, J. S.; and Shah, S. P., Extent of
Healing of Cracked Normal Strength Concrete, Journal of Materials in
Civil Engineering, ASCE, V. 12, No. 1, 2000, pp. 92-96. doi: 10.1061/
(ASCE)0899-1561(2000)12:1(92)
23. Clear, C. A., The Effects of Autogenous Healing upon the Leakage
of Water through Cracks in Concrete, Cement and Concrete Association,
Wexham Springs, UK, 31 pp.
24. Edvardsen, C., Water Permeability and Autogenous Healing of
Cracks in Concrete, ACI Materials Journal, V. 96, No. 6, Nov.-Dec. 1999,
pp. 448-455.

ACI Materials Journal